Friday, September 9, 2011

The Scope of American Student Under-Achievement

A couple of my readers have been suggesting that the problem with American student achievement is actually only a problem with our poorest and lowest-performing students, whose weak scores are dragging down the averages. I don't think that’s the case, and I want to present some data on this question, because it’s an important one.

I’m looking at the 2009 international assessment scores (from the Programme for International Student Assessment, which are available online). Now, test-scores don’t tell you everything, of course, but when you need a blunt comparative instrument, they’re a good one. According to the data from these exams, high-performing US students are indeed more competitive with high-performing students from other countries than low-performing US students are with low-performing students from other countries, but there’s still plenty of room for concern.

The table below compares percentiles of the US population to equivalent percentiles in other nations in the Organisation for Economic Cooperation and Development (OECD). There are 34 nations in the OECD, and the data in each cell of the table shows how the US ranks among those 34 nations for a particular percentile in a particular subject. For example, if we compare the science scores of the 75th percentile in science, for every nation in the OECD, we find the US’s 75th percentile ranking 16th, or just above the middle.


As you can see from the table, our 25th percentile is actually somewhat worse than our 10th, but neither are very competitive. Our higher percentiles are doing pretty well in reading, fair to middling in science, and quite poorly in math. Now, I find all that troubling. Remember that all of those rankings are comparing these percentiles to equivalent percentiles in other countries, so this doesn’t just mean that our 25th percentile isn’t doing that well; it means that our 25th percentile is doing worse than the 25th percentile in most other economically developed nations. If we were doing a bad job of educating 5 or 10 percent of our population, I might call that an isolated problem, but if we’re talking about a quarter of our population—and the entire population in at least one subject—then I’d call that mediocre.

But the situation may actually be worse than those numbers make it look. In PISA’s analysis of their data, they use regressions to identify national characteristics that are correlated with student achievement—GDP per capita, per-pupil educational expenditure, etc. In most of these variables, the US is at the upper end of the spectrum. In other words, based on how wealthy we are and how much we spend on education, our students ought to be scoring well above average. By these measures, the US is the 6th most advantaged nation in the OECD.

To give a sense of what this means, PISA adjusts national reading scores to control for various relevant variables. When all national reading scores are adjusted to control for GDP per capita, the US drops from 13th place to 17th. When reading scores are adjusted to control for per-pupil educational expenditure (but not GDP per capita), the US drops to 20th. PISA’s 2009 analysis focuses on reading, so current data on the effect of these variables on math and science scores is not readily available, but I see no reason to assume the effects would not be similar.

Of course, America has unique educational challenges, stemming primarily from the tremendous cultural diversity of our population. We might excuse our schools if they don’t bring our kids up to the level of Finnish or Korean students. You can call this a demographic problem, but it’s not an isolated problem. It’s not a matter of some small ghettoized minority of our students underperforming.

As usual, I’m not blaming the schools for this problem; I’m not even saying the schools alone can necessarily fix it. But looking at this data, I have to conclude that there’s a need for improvement, not just in the inner-city, but throughout our school system.


  1. We might excuse our schools if they don’t bring our kids up to the level of Finnish or Korean students. You can call this a demographic problem, but it’s not an isolated problem. It’s not a matter of some small ghettoized minority of our students underperforming.

    It's not "small" nor is it "ghettoized", but it is indeed a largely isolated demographic problem.

    If we remove blacks and Hispanics from our population, that the top 25% rises dramatically. The US has a wider range of ability; thus its percentiles are dramatically different. By ignoring race, you overstate the degree of our so-called underperformance.

    The only real way to get a sense of how America is doing is to break it down by race--not because race determines cognitive ability, but because the distributions of cognitive ability are skewed by race, for reasons we don't understand yet.

    And yes, America still underperforms slightly, but that's because our diversity and the political demands that come with it prevent us from tracking by ability. We aren't allowed to educate our top students as well as we could, nor are we allowed to educate our low ability kids in the conditions that might help them achieve better.

    In an earlier post, you said that the data you have access to will prove your point. You have not offered data that is new or surprising.

    Incidentally, my counter--that America does a good job of educating its population, given its demographics--is not particularly controversial outside of eduformer and progressive circles.

  2. As an 18-year career educator in independent schools - four to be exact - working with predominately White, affluent, upper-middle class students, I am living proof that student achievement in the United States is not the exclusive domain of, "the poorest and lowest-performing students."

    The fact of the matter is: No foundation would fund research for a comprehensive study on the low achievement of White, affluent students in the United States; there is no pathology there, i.e. White Americans are not viewed as pathological, as they are the default for high achievement, intelligence, etc. To the contrary, poor, low-performing students, who are often Black and Brown, present the ideal pathology, the ideal caricature, because, Black and Brown people are inherently pathological. and have been presented as such since time immemorial, and still are. So, that fact hasn't changed much.

    That is not to say that the achievement gap amongst poor, low-performing students who are disproportionately Black and Brown should not be studied; it should. It would be better, however, if the 30-odd years of funded research on this population and on this issue actually netted some lasting and enduring change educationally ad economically for these kids. So, perhaps the research funding should be linked to devising an actual solution for instituting lasting and enduring change?

  3. I meant to say, "student achievement *problem* in the United States..."

  4. You aren't "living proof" of that. You may have anecdotal experience that you believe exemplifies the "problem", but you aren't living proof.

    Incidentally, I never mentioned income. For some reason, many people read "low ability" as "low income". I realize that Max mentioned it, but I just wanted to be clear that low income != low ability. Race is, as I said, a much better predictor than income. (For example, low income whites routinely outscore middle to high income blacks on all tests of cognitive ability). But again, this does not mean that race DETERMINES ability. It just means that,given the strong ability skews by race, we can't meaningfully assess our educational system unless we compare like to like on race.

    There is no evidence that our top students (regardless of affluence level) have a problem, apart from the one I mentioned--that we could be teaching them more, if we separated by ability. Your anecdata involves, apparently, *wealthy* white students, but I was talking about ability, regardless of income or race--and there are high ability students in all races. The issue is frequency of distribution.

    So, perhaps the research funding should be linked to devising an actual solution for instituting lasting and enduring change?

    Notice all the misconceptions you have. First, you equate low income with low ability. Second, you assume a solution is possible. Third, you predefine the solution--low income students performing at the level of high income students.

    There's just so much wrong there. First, while low income students are disproportionately low ability, there are high ability students in low income popuations--and I am sure we are losing some of them. How many, we do't know.

    Second, you assume that a solution is possible, and you know what the solution must be. But never in the history of educational research have we demonstrated that low ABILITY students can be taught to the same level as high ABILITY students. Since everyone tends to conflate income and ability, they are eager to point out a case in which low income students were, at least temporarily, cherry picked and test selected into performing better (although rarely equal). But again, the problem is ability, not income (although income complicates the issue).

    There is zero evidence that low cognitive ability students can be taught to the same level as high cognitive ability students. So we can't "devise an actual solution". None exists, and our current knowledge would argue that none is possible.

  5. Cal, you've made a highly controversial, potentially offensive switch from a discussion about student achievement to one about student ability. Obviously, state test data tells us nothing conclusive about ability. You've referred to tests of cognitive ability, so I assume you're looking at data on the subject. If you insist on opening this kind of discussion, please provide a link or a citation or something.

  6. I should clarify that there are developmental windows in childhood, and kids who do not receive proper physical and psychological nourishment during these windows may come out cognitively underdeveloped and ill-equipped to attain high achievement levels. We can call this student ability, but we should be careful to clarify that this is not necessarily innate ability.

    The above problem is the basis for my position that early-childhood and prenatal interventions must be a major part of any education reform movement aimed at reducing or eliminating the achievement gap. Such considerations, however, hardly obviate the need for educational improvement at the pre-primary, primary, and secondary levels.

  7. @Cal - should I have said, "dead proof?"

  8. Max, are you saying it's controversial to link achievement to ability? I wasn't aware of any such controversy. I agree that high ability students can have low test scores and low incentives, but it's largely impossible for students who achieve a great deal to have low ability.

    So far as my statement goes, I deliberately stayed away from cause entirely. I do not care, for purposes of this discussion, why a student is low ability when he or she reaches kindergarten, and why this level of ability generally (but not always) stays constantly low. I don't know if it's innate, nature, nurture, or whatever. It doesn't matter to this point, here:

    But never in the history of educational research have we demonstrated that low ABILITY students can be taught to the same level as high ABILITY students.

    I think this is generally indisputable. Two fifth graders, one of who reads at a first grade level, one reads at an 8th grade level--we have never been able to prove that the first one can learn to the same level as the second. Two eighth graders, one of whom has fourth grade level math skills, one who understands proportional thinking and is fluent in math facts--no one has ever demonstrated a method for teaching them to the same level.

    If you have research proving otherwise, great. But I think that statement is pretty indisputable. So I'm not sure why you think anything I've said is controversial.